Energies: Oscillation Suppression Method by Two Notch Filters For Parallel Inverters Under Weak Grid Conditions

energies
Article
Oscillation Suppression Method by Two Notch Filters
for Parallel Inverters under Weak Grid Conditions
Ling Yang 1 , Yandong Chen 1, * , Hongliang Wang 1 , An Luo 1 and Kunshan Huai 2
1 College of Electrical and Information Engineering, Hunan University, Changsha 410082, China;
yangling_1992@163.com (L.Y.); liangliang-930@163.com (H.W.); an_luo@126.com (A.L.)
2 Guangzhou Power Supply Co., Ltd., Guangzhou 510620, China; huaikunshan@126.com
* Correspondence: yandong_chen@hnu.edu.cn; Tel.: +86-151-1626-8089

Received: 4 November 2018; Accepted: 5 December 2018; Published: 8 December 2018
Abstract: With plenty of parallel inverters connected to a weak grid at the point of common coupling
(PCC), the impedance coupling interactions between the inverters and the grid are enhanced, which
may cause high-frequency harmonic oscillation and further aggravate the system instability. In this
paper, a basic technique for inverter output impedance is proposed to suppress the oscillation,
showing that the inverter output impedance should be designed relatively high at the harmonic
oscillation frequency, while relatively low at other frequencies. On the basis of the proposed technique,
two virtual impedances are added to be in parallel and in series with the original inverter output
impedance, respectively. Thus, an oscillation suppression method by two notch filters is proposed to
realize the virtual impedances and increase the whole system damping. The implementation forms
of the virtual impedances are presented by the proposed PCC voltage feedforward and grid-side
inductor current feedback with two notch filters. Finally, simulation and experimental results are
provided to verify the validity of the proposed control method.
Keywords: weak grid; parallel inverters; oscillation suppression; notch filter; impedance reshaping
1. Introduction
With the increasing energy crisis and environmental problems, renewable energy generation,
mostly in power plants or microgrids, have grown rapidly [1–3]. Due to the distributed locations of
renewable energy generators, multiple transformers and long transmission lines are utilized to connect
the systems to the public grid [4]. Thus, the public grid exhibits the feature of a weak grid in which
grid impedance cannot be ignored [5].
Especially, in a large-scale power plant or microgrid, renewable energies are mostly connected
to the grid via parallel inverters [6,7]. By this way, the power plant or microgrid can easily expand
the output power capacity, and it is also convenient to connect plenty of renewable energies into the
grid [8]. Under weak grid conditions, the impedance coupling interactions between the inverters and
the grid are enhanced further due to grid impedance [9]. If control parameters and device selection of
all inverters are the same, the grid impedance seen by each inverter will become n times of the real
grid impedance [10]. Therefore, these interactions may cause oscillation if inverters are improperly
designed or controlled.
There are two methods to analyze the system’s stability. The first is the eigenvalue-based
analysis [11], which is usually utilized to evaluate the system’s stability. The eigenvalue-based stability
analysis studies the eigenvalues of a system’s state space model matrix, which requires the physical
features and control parameters in the system [12]. The second one is the impedance-based stability
criteria [13,14], which is well built to adjudicate the system stability. The system will be stable if two
requirements are satisfied [13]. The first requirement is that the grid-connected inverter is stable in
Energies 2018, 11, 3441; doi:10.3390/en11123441 www.mdpi.com/journal/energies

Energies 2018, 11, 3441 2 of 20
the public grid, and the other is that the product of grid impedance and inverter output admittance
satisfies the Nyquist criterion. The industrial and academia community have widely accepted the
theory of impedance-based stability analysis [15,16].
Some strategies were proposed to suppress the oscillation, which can be divided into two cases.
One is to introduce additional hardware equipment to the PCC [17], the other is to reshape the inverter
output impedance [18–23]. Adding additional equipment was adopted to stabilize the paralleled
multi-inverter system, in which structures and control parameters of the inverters are unknown.
Reference [17] installed an active damper at the PCC to suppress the oscillation. However, extra cost
may exist owing to the demand of additional circuitry [24].
Different impedance reshaping methods need to be adopted to suppress the oscillation. The
existing impedance reshaping methods mainly include: capacitance-current-feedback methods [18,19],
capacitance-voltage-feedback methods [20,21], and virtual resistor methods [22,23]. In Reference [18],
the real-time computational method was proposed to decrease the computational delay, which can
simplify the design and enhance the performance. However, if the grid voltage has much harmonics,
it will make the duty cycle of the inverters change sharply. Similar to the capacitance-current-feedback
methods, useful active damping was induced by the derivative feedback of the capacitance
voltage [20,21]. However, the capacitance-voltage-feedback methods should handle the challenge
of grid voltage variation. In References [22,23], the virtual resistor methods made the inverter
output impedance show high impedance at all frequencies, which can suppress the oscillation.
However, due to the high impedance characteristics of the inverters at the fundamental frequency,
the methods will cause the change of the fundamental current, and then influence the tracking
precision of the grid-connected power. Therefore, the fundamental impedance and high-frequency
harmonic impedance of inverters are separately considered. The fundamental impedance should be
designed as the low impedance, which does not affect the grid-connected power tracking. However,
the high-frequency harmonic impedance is designed as the high impedance to suppress the oscillation.
In this paper, the oscillation suppression method by two notch filters is proposed to increase
the whole system damping. This paper is organized as follows. Section 2 analyzes the oscillation
mechanism of a paralleled multi-inverter system. Section 3 presents the demand for the inverter
output impedance for the purpose of oscillation suppression, and the method of adding the virtual
impedances to meet the demand is proposed, which introduces two notch filters to the PCC voltage
feedforward and grid-side inductor current feedback. Section 4 compares and analyzes the system’s
stability in two cases. Sections 5 and 6 provide the simulation and experimental results to verify the
validity of the proposed control method. Finally, Section 7 gives the conclusion.
2. Oscillation Mechanism of Paralleled Multi-Inverter System
2.1. System Description

The structure of a paralleled multi-inverter system is shown in Figure 1. The left and right side
are the inverter subsystem and the grid subsystem. j = 1, 2, . . . , n. Udc is the DC voltage. uinvj , uC1j ,
and uPCC are the inverter output voltage, filter capacitor voltage, and PCC voltage. ug is the grid
voltage. Zg is the grid impedance. Inductor-capacitor-inductor-type (LCL-type) filter is constituted by
the inverter-side inductor L1j , grid-side inductor L2j , and filter capacitor C1j . RL1j and RL2j are parasitic
resistances of L1j and L2j . iL1j is the inverter-side inductor current. iC1j is the filter capacitor current.
ioj is the grid-side inductor current. ig is the grid-connected current.
Energies 2018, 11, x FOR PEER REVIEW 3 of 20
Energies 2018, 11, 3441 3 of 20
Figure 1. Structure of paralleled multi-inverter system.

Figure 1. Structure of paralleled multi-inverter system.
2.2. Oscillation Mechanism Figure 1. Structure of paralleled multi-inverter system.
2.2. Oscillation Mechanism
In the grid-connected
2.2. Oscillation Mechanism mode, single inverter is equivalent to the current source ij in parallel with
In the admittance
equivalent grid-connected mode, is
Yj, which single inverterequivalent
the Norton is equivalent to the
circuit [23]. current
The grid source ij in paralleltowith
is equivalent the
In the grid-connected mode, single inverter is equivalent to the current source ij in parallel with
equivalent
grid voltage admittance Yj , with
ug in series whichgrid is the Norton equivalent
impedance Zg. Fromcircuit the PCC, [23].the TheNorton
grid is equivalent
equivalenttocircuit
the gridof
equivalent admittance Yj, which is the Norton equivalent circuit [23]. The grid is equivalent to the
voltage ug multi-inverter
paralleled in series with grid
systemimpedance
is shown Zgin
. From
Figurethe2a.PCC,From the Norton
Figure 2a,equivalent
the circuitcircuit of paralleled
relationship about
grid voltage ug in series with grid impedance Zg. From the PCC, the Norton equivalent circuit of
umulti-inverter
PCC is shown in system is shown
Equation (1) by the in Figure
nodal 2a. Figure 2a, the circuit relationship about uPCC is
Frommethod.
analysis
paralleled multi-inverter system is shown in Figure 2a. From Figure 2a, the circuit relationship about
shown in Equation (1) by the nodal analysis method.
uPCC is shown in Equation (Y1(1)+ Y2by + + Ynodal
the n )u PCC analysis g = (i1 + i2 +  + in ) + u g Z g
+ u PCC Z method.
(1)
(Y1 + Y(2(YY+
1++ 
YY·2 ·+++
·  Y++nYY)nu))PCC
uuPCC+++ uuPCC
PCC ZZg g===
/Z ((ii1(++i1ii+
2++i
2 ++
+ ii·n ·))·+
++uugin )Z
Z+g ug /Zg (1)
1 2 n PCC PCC g 1 2 n g g
(1)
where (Y1 +Y2 +of a+Ysingle n )u + uPCC Z gand = (i Y+j iis +the in ) + ug Z g admittance of a single
 +equivalent
where iij j isisthe
thecurrent
current source
source of a single PCCinverter,
inverter, and Yj is 1the 2equivalent admittance of a single inverter.
inverter.
where ij is the current source of a single inverter, and Yj is the equivalent admittance n
of a single
inverter. ig =  ioj Z g
j =1
n
ig =  ioj Zg
n n
j =1 +
i j Y j
uPCC ug
j =1
n
j =1
n
+-
 i j source
current
j =1
 Yj
equivalent
j =1
admittance
uPCC ug
-
(b)
current source equivalent admittance
n
igh =  iohj (b)
n
igh =  iohj j =1
Zgh
j =1 n
igh =  iohj
n
igh =  iohj j =1
Zgh
j =1
n series
Y
n n
 ihj
j =1
 Yhj
j =1 j =1
hj
series
ugh
n harmonic
Y
n n
i
j =1
hj Y
j =1
hj
j =1
hj voltageugh
harmonic
(d)voltage
Figure 2.
Figure Output admittance
2. Output admittance model
model and
and oscillation
oscillation mechanism
mechanism of of paralleled
paralleled multi-inverter system.
(d) multi-inverter system.
(a) Norton equivalent circuit; (b) Output admittance model; (c) Parallel oscillation; (d) Series
(a) Norton equivalent circuit; (b) Output admittance model; (c) Parallel oscillation; (d) Series oscillation.
Figure 2. Output admittance model and oscillation mechanism of paralleled multi-inverter system.
oscillation.
(a) Norton equivalent circuit; (b) Output admittance model; (c) Parallel oscillation; (d) Series
oscillation.
Energies 2018, 11, 3441 4 of 20
From Equation (1), the output admittance model of paralleled multi-inverter system satisfies the
Equation (2), as depicted in Figure 2b.
n n
uPCC ∑ Yj + uPCC /Zg = ∑ i j + ug /Zg (2)
j =1 j =1
The frequency of harmonic current ihj is caused by the system nonlinear factor. If it is equal to or
close to the parallel resonance frequency of impedance network, it will cause the parallel oscillation,
as shown in Figure 2c. The frequency of harmonic voltage ugh is caused by the grid distortion. If it is
equal to or close to the series resonance frequency of impedance network, it will result in the series
oscillation, as shown in Figure 2d.
From Figure 2b, the grid-connected current ig can be derived as
n n
1
ig = ( ∑ i j − ug ∑ Yj )·( n ) (3)
j =1 j =1 1 + Zg · ∑ Yj
j =1
where the product of grid impedance and inverter output admittance is defined as the impedance ratio
K, which can be expressed as
n
K = Zg · ∑ Yj (4)
j =1
From Equation (3), it can be assumed that the grid voltage is stable in the absence of the inverter,
and the inverter will be stable if the grid impedance is zero. Nevertheless, when the grid impedance is
not negligible, the system will be stable only if the impedance ratio K satisfies the Nyquist criterion in
Reference [12]. In other words, the system will be stable, only if the Nyquist curve of the impedance
ratio does not surround (−1, j0).
Thus, there are impedance coupling interactions between the inverters and the grid, which will
aggravate the harmonic distortion of grid-connected current, lead to the oscillation in paralleled
multi-inverter system, and even cause the system to be unstable.
3. Oscillation Suppression Method by Two Notch Filters for Parallel Inverters
3.1. Demand for the Inverter Output Impedance

To suppress the oscillation, the inverter output impedance should be designed relatively high
at the harmonic oscillation frequency while relatively low at other frequencies. For this purpose,
the virtual impedances are added to be connected with the original output impedance, as shown in
Figure 3. Figure 3a–c presents parallel, series, and parallel-series virtual impedances, respectively.
The above three forms can reach the same performance for oscillation suppression. For the first and
second forms, it is relatively complicated to introduce one virtual impedance to realize that the inverter
output impedance shows high at the harmonic oscillation frequency while relatively low at other
frequencies. However, the third form with two virtual impedances is relatively easy to achieve this
purpose. Thus, the third form is mainly discussed in this paper.
Energies 2018, 11, 3441 5 of 20
Z
n n
ig =  ioj ig =  ioj ssj
j =1 ifj if3j j =1 j =1 Zg
ihj ih3j
n n
ih1j if1j n
n n
i
j =1
ooj
n
i
j =1
pj
n n
i
j =1
ooj
grid equivalent
impedance Zsgj
 ij
j =1
 Z oj
j =1
 Z ppj
j =1
 ijj =1
 Z oj
j =1
ug
self-impedance equivalent
impedance
(b)
n
Z
n
ig =  ioj sj
j =1 j =1
n
n n
Z
j =1
sj
n n
i
j =1
ooj
n
i
j =1
pj
i
j =1
j Z
j =1
oj Z
j =1
pj
Figure 3. Equivalent schematic diagram.

diagram. (a)
(a) Parallel virtual impedance;
impedance; (b) series virtual impedance;
impedance;
(c) parallel-series virtual impedances; (d) equivalent
equivalent output
output impedance
impedance model.
model.
In Figure
In Figure3c, 3c,ZojZojis isthethe
self-impedance
self-impedance when whenthe parallel-series
the parallel-series virtual impedances
virtual are notare
impedances added.
not
The equivalent impedance Z (Z = 1/Y ) is formed by Z in parallel
added. The equivalent impedance Zj (Zj = 1/Yj) is formed by Zoj in parallel with Zpj, and then in series
j j j oj with Z pj , and then in series with Zsj .
The grid equivalent impedance
with Zsj. The grid equivalent impedance Z sgj consists of the virtual impedance Z and the
Zsgj consists of the virtual sjimpedance Zsj and the gridgrid impedance Zg
connected inZseries.
impedance if/hj , if/h1j
g connected
if/h2j , and
in ,series. if/hj,if/h3j
if/h1j,are total
if/h2j , and fundamental/high-frequency harmonic current,
if/h3j are total fundamental/high-frequency
fundamental/high-frequency
harmonic current, harmonic current
fundamental/high-frequency of Z oj branch,
harmonicfundamental/high-frequency
current of Zoj harmonic
branch,
current of Z branch,
fundamental/high-frequency
pj and fundamental/high-frequency harmonic current
harmonic current of Zpj branch, and fundamental/high-frequency of Z sgj branch, respectively.
if/hj = if/h1jcurrent
harmonic + if/h2jof +Z if/h3j. The total fundamental frequency current ifj and
sgj branch, respectively. if/hj = if/h1j + if/h2j + if/h3j. The totalthe total high-frequency
fundamental frequency
harmonic current i are determined by the shunt circuit,
current ifj and thehjtotal high-frequency harmonic current ihj are determined by the shuntconsisting of the self-impedance Z oj , parallel
circuit,
virtual impedance Z
consisting of the self-impedance
pj , and grid equivalent impedance
Zoj, parallel virtual impedance Z sgj . From the PCC, the Norton
Zpj, and grid equivalent impedance equivalent
circuit
Z of Figure 3c is refined into the forms of Figure 3d.
sgj. From the PCC, the Norton equivalent circuit of Figure 3c is refined into the forms of Figure 3d.
On the basis
On the basis of of thethe proposed
proposedbasic basictechnique,
technique,the theparallel
parallelvirtual
virtual impedance
impedance ZpjZshould
pj should be
be designed
designed to toshowshow low lowimpedance
impedanceatatthe theharmonic
harmonic oscillation
oscillation frequency.
frequency. By By doing
doing so, so, most
most
high-frequency harmonic
high-frequency harmoniccurrent current willwill
flowflow into theintoparallel virtual impedance
the parallel Zpj branch;
virtual impedance Zpjit branch;
effectively it
suppresses the oscillation of paralleled multi-inverter system.
effectively suppresses the oscillation of paralleled multi-inverter system. In the meantime, In the meantime, to improve the powerto
quality
improveofthe grid-connected
power quality current, the series virtual
of grid-connected impedance
current, the series Zsj should
virtualbeimpedance
designed to Zsjdisplay
shouldlow be
impedance at the fundamental frequency. This way, most
designed to display low impedance at the fundamental frequency. This way, most fundamental fundamental frequency current flows into
the grid branch
frequency currentwith relatively
flows into the low gridimpedance.
branch with relatively low impedance.
3.2. Oscillation Suppression Method

3.2. Oscillation Suppression Method
The parallel-series virtual impedances in Figure 4 can be realized by introducing two notch filters.
The parallel-series virtual impedances in Figure 4 can be realized by introducing two notch
In Figure 4, two virtual impedances are added in parallel and series with inverter output impedance,
filters. In Figure 4, two virtual impedances are added in parallel and series with inverter output
respectively. Gi is the grid-connected current loop proportional resonant (PR) controller, GPWM is the
impedance, respectively. Gi is the grid-connected current loop proportional resonant (PR) controller,
equivalent gain of the inverter, ZL1j = sL1j + RL1j , ZC1j = 1/sC1j , ZL2j = sL2j + RL2j .
GPWM is the equivalent gain of the inverter, ZL1j = sL1j + RL1j, ZC1j = 1/sC1j, ZL2j = sL2j + RL2j.
The parallel virtual impedance Z and series virtual impedance Zsj can be expressed as
The parallel virtual impedance Zpj pj and series virtual impedance Zsj can be expressed as
ZZpjpj == rr11//G
GNN
(
 (5)
(5)
= r22GGNN
ZZsjsj =
where r11 and r22 are

are the
the proportional
proportional coefficient
coefficient and GNN isisthe
thenotch
notch filter.
filter.
The grid-connected current loop PR controller
controller G
Gii can be expressed as
2k ω s
GG + 2 2ki1i1 ωcc s 2
i =kkpp + (6)
i = ωccss ++ωωo o2
+ 22ω (6)
ss2 +
Energies 2018, 11, 3441 6 of 20
where kp is the proportional coefficient of quasi-proportional resonant controller, ki1 is the resonance
where kp is the proportional coefficient of quasi-proportional resonant controller, ki1 is the resonance
gain of quasi proportional resonance controller, ω c is the cut-off angular frequency, and ω o is the
gain of quasi proportional resonance controller, ωc is the cut-off angular frequency, and ωo is the
fundamental angular frequency.
fundamental angular frequency.
The effects of dead-time of switching devices in paralleled multi-inverter system are regarded
The effects of dead-time of switching devices in paralleled multi-inverter system are regarded
as a disturbance, which have a constant amplitude and an alternative direction depending on the
as a disturbance, which have a constant amplitude and an alternative direction depending on the
inverter-side inductor current iL1j [25]. It is notable that the disturbance can be seen as the controlled
inverter-side inductor current iL1j [25]. It is notable that the disturbance can be seen as the controlled
current source i in Norton equivalent circuit, which can be presented as
current source idjdjin Norton equivalent circuit, which can be presented as
idji = UUejej ·signi(iL1j ) (7)

dj = Zpj A + Zsj B ⋅ sign( L1j ) (7)
Z pj A + Z sj B
where AA = L1jZZL2j++
= ZZL1 ZC1j ( Gi GPWM + ZL1j + ZL2j ) and B = ZL1j ZC1j + ( ZL1j + ZC1j )( ZL2j + Zpj ).
where j L2j Z C1j (Gi GPWM + Z L1j + Z L2j ) and B = Z L1j Z C1j + ( Z L1j + Z C1j )( Z L2j + Z pj ) .
According to Reference [25], Uej in Equation (7) can be expressed as
According to Reference [25], Uej in Equation (7) can be expressed as
ttdj +
+ ttononj − toffj
Uej U= 2=(U Dj −−U
+ +UU Tj )) dj j − toffj − U−DjU− UTj (8)
ej 2(U
dcj
dcj Dj U Tj T −U Dj Tj (8)
Tsjsj
where U and U are the on-state voltage drop of switching devices and diodes, T , t , tonj and toffj
where UTjTj and UDjDjare the on-state voltage drop of switching devices and diodes, Tsjsj, dj
tdj, tonj and toffj
are switching period, dead-time, turn-on time, and turn-off time of switching devices.
are switching period, dead-time, turn-on time, and turn-off time of switching devices.
Meanwhile, sign(iL1j ) in Equation (7) can be expressed as
Meanwhile, sign(iL1j ) in Equation (7) can be expressed as
> 0> 0
11 iL1ijL1j
(
sign(
sign )=
(iL1jiL1) j = (9)
(9)
 −1 iL1
− 1 i j <0
L1j <0
From
From Figure
Figure 4,
4, the
the closed-loop
closed-loop transfer function of the system can be expressed as
Gi GGPWM Z C1jZZC1j
i GPWM pj Zpj C1j +Z( Z L1+
Z L1j ZZ j +( Z
ZL1j
C1j )( j +)(
+ZZpC1j ZZ j ) + ZL2j )
L2pj
iioojj ==GGj ij refj ++
irefj idij dj− −
YjY
uPCC =
j uPCC =
irefj +irefj − idj − L1j C1j
idj + uPCC uPCC(10)
(10)
Z pj AZpj+AZs+j BZsj B Z pj A +ZZpjsjAB + Zsj B
where
where iirefj
refjisisthe
thereference
referencecurrent
currentof
ofsingle
singleinverter,
inverter, G
Gjj isisthe
the current
current source
source equivalent
equivalent coefficient
coefficient of
of
single
single inverter,
inverter, and Yjj isis the
the equivalent
equivalent admittance
admittance of
of single inverter.
Figure 4. Equivalent control block diagram of oscillation suppression method by two notch filters for
Figure 4. Equivalent control block diagram of oscillation suppression method by two notch filters for
parallel inverters.
parallel inverters.
According to Equation (10), the refined equivalent output impedance model is shown in Figure 5.
According to Equation (10), the refined equivalent output impedance model is shown in Figure
The current source ij is equivalent to the current source i1j in parallel with the current source idj .
5. The current source ij is equivalent to the current source i1j in parallel with the current source idj.
Two virtual impedances are added to be in parallel and in series with the original inverter output
Two virtual impedances are added to be in parallel and in series with the original inverter output
impedance, respectively. Thus, Figure 5 is equivalent to the refinement of Figure 3d, which can achieve
impedance, respectively. Thus, Figure 5 is equivalent to the refinement of Figure 3d, which can
the proposed approach.
achieve the proposed approach.
Energies 2018, 11, 3441 7 of 20
Z
j =1
sj
Figure 5. Refined equivalent output impedance model.
From
From Figure
Figure 5,
5, the
the equivalent
equivalent model
model of
of the
the m-th
m-th inverter
inverter can
can be
be expressed
expressedas
as
= G=
iomiom Gm i+refm
m irefm − Yidm
idm + m uPCC
− Ym uPCC
Gi Gi PWM
G G PWM Z C1m
Z C1mZ pm
Zpm Z Z Z ZC1m++(( ZZL1m +ZZ
L1m + )(pm
)( Z Z pm + Z)L2m )
+ ZL2m (11)
(11)
== Zpm A+Zsm B iirefm + idm −L1mL1mC1m C1m
C1m uPCCu
refm + idm − Zpm A+ Zsm B PCC
Z pm A + Z sm B Z pm A + Z sm B
where irefm , Gm , Ym , idm , ZL1m , ZC1m , ZL2m , Zpm , and Zsm are the variables of the m-th inverter.
where irefm, Gm, Equation
Substitute Ym, idm, ZL1m(1), into
ZC1mEquation
, ZL2m, Zpm(11),
, andand
ZsmEquation
are the variables
(11) can of
bethe m-th inverter.
rewritten as
Substitute Equation (1) into Equation (11), and Equation (11) can be rewritten as
n
iom = Gselfm irefm + n ∑ G paralm, j ire f j − Gserim ug + idm
iom = Gselfm irefm + j=1,jG
6=paralm,
m j irefj − Gserim ug + idm
Z Yj =1,Gj ≠ m n Z Y Ym
(12)
= ( Gm − Z Yg mGn m )irefm + n ∑ (− Z Yg nm Gj )ire f j − ug + idm
1+YZmg ∑ Yj
n
(12)
=(Gm − 1+Zg jn=∑1 Yj ) irefm + j
g m m g m
=1,j6=(m − 1+Zgn j=∑1 YG j
j )irefj − jn=1
ug + idm
1 + Zg  Y j j =1, j ≠ m
1 + Zg  Y j 1 + Zg  Y j
where Gselfm is the transfer relationship between the grid-side
j =1 j =1 inductance j current =1 iom of the m-th
inverter and reference current irefm of the m-th inverter, Gparalm,j is the transfer relationship between
where Gselfm is the transfer relationship between the grid-side inductance current iom of the m-th
the grid-side inductance current iom of the m-th inverter and reference current irefj of the jth inverter,
inverter and reference current irefm of the m-th inverter, Gparalm,j is the transfer relationship between
and Gserim is the
the grid-side transfer current
inductance relationship iom ofbetween
the m-thgrid-side inverter and inductance
reference current
current iomirefj
ofofthethem-th inverter
jth inverter,
and grid voltage u g
and Gserim is the transfer. relationship between grid-side inductance current iom of the m-th inverter
When
and grid a similar
voltage ug. analysis method is adopted to other inverters, the system can be expressed using
a closed-loop transferanalysis
When a similar functionmethod matrix with
is adopted reference currents,
to other grid voltage
inverters, the system and controlled current
can be expressed
sources as inputs and grid-side inductor currents as outputs
using a closed-loop transfer function matrix with reference currents, grid voltage and controlled
current
 sources as inputs and grid-side inductor currents  as outputs     
io1 Gself1 Gparal1, 2 · · · Gparal1, n iref1 Gseri1 id1
 io2   Gparal2,   Gself1 Gself2 Gparal1, paral2, n n iiref2
Gparal1, ref1   G seri1Gseri2   d1   id2 
  i i

· · ·2 G     
 .  =   o1 . 1   
.  G 
−  ..  u  + (13)
 .   io2 . G . Gself2..  .. Gparal2,n iref2 id2g  .. 
      
 .    . =  paral2, 1.. . .   ..   −  seri2  u. g +     .  (13)
ion  1 Gparaln, 2 · · · Gselfn   irefn   Gserin  
Gparaln, idn
i   G G  G
  
i G  i 
 on   paraln, 1 paraln, 2 selfn    refn   serin   dn 
It is obvious that the strong coupling between the inverters and the grid may exist and introduce
It is oscillation
harmonic obvious that the strong
currents under weak coupling grid between
condition.the inverters and the grid may exist and
introduce harmonic oscillation currents
According to Figure 4, Figure 6 gives the control under weak gridblockcondition.
diagram of the oscillation suppression
method According
by twotonotchFigure 4, Figure
filters 6 gives inverters.
for parallel the control The blockfirst diagramnotch of the oscillation
filter is introduced suppression
into the
method by two
feedforward pathnotch
of PCCfiltersvoltage forand parallel
the second inverters. notchThe filter first notch filterinto
is introduced is the
introduced
feedbackinto paththe
of
feedforward path of PCC voltage and the second notch filter is
the grid-side inductor current. The feedback path of the grid-side inductor current with the notch filterintroduced into the feedback path of
theequivalent
is grid-side to inductor
a virtual current.
impedance The feedback
in series with pathinverter
of the grid-side
output impedance,inductor current which with the notch
can effectively
filter is equivalent to a virtual impedance in series with
improve the power quality of the grid-connected current. The feedforward path of PCC voltage with inverter output impedance, which can
effectively
the improve
notch filter equals thetopower
a virtual quality of the grid-connected
impedance in parallel withcurrent. inverterThe outputfeedforward
impedance. pathThis
of PCC
can
voltage with
effectively the notch
restrain the parallelfilter inverters’
equals toharmonic a virtualcurrent impedance from flowing in parallel into thewith gridinverter
and avoid output
the
impedance. This can effectively restrain the parallel inverters’
oscillation phenomenon. H1j is the feedback coefficient of the grid-side inductor current, and H2j is the harmonic current from flowing into
the grid andcoefficient
feedforward avoid theofoscillation
PCC voltage. phenomenon. H1j is the feedback coefficient of the grid-side
inductor current, and H2j is the feedforward coefficient of PCC voltage.
Energies 2018, 11, 3441 8 of 20
Figure 6.6. Control block diagram

Figure diagram of the
the oscillation
oscillation suppression
suppression method
method by
by two
two notch
notch filters
filters for
for
parallel inverters.
parallel inverters.
From
FromFigure
Figure6,6,the
theequivalent
equivalentclosed-loop
closed-looptransfer
transferfunction
functionofofthe
thesystem
systemcan
canbe
beexpressed
expressedasas
Gi GGPWM
G ZC1j
Z j ZL1j + j + H 2 j2j
+ZZC1C1j + H GG PWMZ C1
ZjC1j
= G+jeqidj
ioj = Gjeqioij refj j + ijeq
iref− Ydj −uY = = i PWM C1irefj
jeq uPCC C
PCC iref+ i idj−− L1j
j +dj
PWM
uPCC
uPCC (14)
(14)
+CD+ +
D+E
E C+D+E
C + D + E
whereGGj eq
where jeq is the equivalent coefficient of the current current source
source after
after the
the single
single inverter
inverter transformation,
transformation,
YYj eq
jeq is the inverter
inverter equivalent
equivalent admittance
admittance after afterthethesingle
singleinverter transformation,C C==ZZL1j
invertertransformation, Z L2j ,,
L1j L2j
D = (1 + H1j ) Gi GPWM ZC1j , and E = ZC1j ( ZL1j + ZL2j ).
D = (1 + H1 j )Gi GPWM Z C1j , and E = Z C1j ( Z L1j + Z L2j ) .
In order to achieve the same purpose of Figures 4 and 6, the equivalent coefficient of the current
source Inand
order thetoinverter
achieve equivalent
the same purpose of Figures
admittance 4 and 6,(10)
in Equation the are
equivalent
equal tocoefficient of the current
those in Equation (14).
sourceit and
Thus, can be theexpressed
inverter equivalent
as admittance in Equation (10) are equal to those in Equation (14).
Thus, it can be expressed as
 GG ZC1j Zpj Gi GPWM ZC1j
 i ZPWM ZsjpjB G=
 Gi GPWM ZAC1+j Z
pj i GPWM Z C1j
C+D+E
  ZL1j ZC1j +(ZL1j = +ZC1j )(Zpj +ZL2j ) Z + Z + H2 GPWM ZC1j
(15)
 Z pj A + Zsj B Zpj A+CZsj+BD + E = L1j C1j C+D+E
 (15)
 Z L1j Z C1j + ( Z L1j + ZC1j )( Z pj + Z L2j ) Z L1j + ZC1j + H 2GPWM Z C1j
From Equation (15), the feedback coefficient of =the grid-side inductor current H1j and the
Z pj A + Z s j B
 of PCC voltage C+D+E
feedforward coefficient H2j can be expressed as
From Equation (15), the  feedback coefficient of the grid-side inductor current H1j and the
Zsj ( ZL1j ZC1j +( ZL1j + ZC1j )( ZL2j + Zpj ))
feedforward coefficient of PCC
 H1j = H
 voltage 2j can beGi Gexpressed
PWM ZC1j Zpj
as
Z Z +( Z + Z ) Z
(16)
 H2j Z=sj ( Z L1j ZGC1PWM+ + +
L1j C1j L1j C1j L2j

 j ( Z
Z L1Zj Z C1j )( Z L2j Z pj ))
 H1j =
C1j pj
 Gi GPWM Z C1j Z pj
At the specific frequency, the amplitude of the notch filter is greatly attenuated while the amplitude (16)
Z L1j Z C1j + ( Z L1j + Z C1j ) Z L2j
at other frequencies is almost non-destructive.
H2 j = The notch filter GN can be expressed as
 GPWM ZC1j Z pj

( 2πs fo )2 + 1
At the specific frequency, the G amplitude
N = of the notch filter is greatly attenuated while(17) the
s 2 s
(
amplitude at other frequencies is almost non-destructive.
2π f o ) + + 1
Q2π f oThe notch filter GN can be expressed as
where f o is the fundamental frequency and Q is the s 2

( quality
) + 1 factor of the notch filter.
2π f
The analysis diagrams of the notch filter witho Q = 0.25, 0.5, and 1 are shown in Figure 7.
GN = (17)
From Figure 7a, the larger Q is, the better the notch s 2characteristic
s of the notch filter, but the worse the
( ) + +1
frequency adaptability. From Figure 7b, when Q 2π=fo0.25, Q
the fo
2πcharacteristic equation has two unequal real
poles on the negative real axis of the s plane, which is over-damping. When Q = 0.5, the characteristic
where fo is
equation the
has fundamental
two frequency
equal real poles on the and Q is thereal
negative quality
axis offactor
the sofplane,
the notch
whichfilter.
is critical-damping.
The analysis diagrams of the notch filter with Q = 0.25, 0.5, and 1
When Q = 1, the characteristic equation has the conjugate complex poles on the left half are shown in plane,
Figurewhich
7. Fromis
Figure 7a, the larger
under-damping. FromQ is, the7c,better
Figure the notch
the tuning time characteristic
of notch filter ofwiththeQnotch filter,
= 0.5 is betterbut theQworse
than = 0.25,the
1.
frequency considering
Therefore, adaptability.the
From Figure
notch 7b, when Qand
characteristics = 0.25, the characteristic
dynamics equationthe
comprehensively, hasQtwo unequal
value was
real poles
selected on0.5.
to be the negative real axis of the s plane, which is over-damping. When Q = 0.5, the
characteristic equation has two equal real poles on the negative real axis of the s plane, which is
critical-damping. When Q = 1, the characteristic equation has the conjugate complex poles on the left
half plane, which is under-damping. From Figure 7c, the tuning time of notch filter with Q = 0.5 is
better than Q = 0.25, 1. Therefore, considering the notch characteristics and dynamics
comprehensively,
half plane, which theisQunder-damping.
value was selected
FromtoFigure
be 0.5.
7c, the tuning time of notch filter with Q = 0.5 is
better than Q = 0.25, 1. Therefore, considering the notch characteristics and dynamics
Energies 2018, 11, 3441 9 of 20
comprehensively, the Q value was selected to be 0.5.
Figure 7. Analysis diagrams of notch filter with Q = 0.25, 0.5, 1. (a) The Bode diagram of notch filter;
Figure 7. Analysis
(b) Pole-zero diagrams
plot of notch filter;of(c)
notch filter
Unit stepwith Q == 0.25,
dynamic
with Q 0.25, 0.5,1.
1. (a)
responses
0.5, (a)of
The
The Bodefilter.
notch
Bode diagram of
diagram of notch
notch filter;
(b) Pole-zero
Pole-zero plot
plot of
of notch
notch filter;
filter; (c)
(c) Unit
Unit step
step dynamic
dynamic responses
responsesof
ofnotch
notchfilter.
filter.
The Bode diagrams of parallel virtual impedance Zpj and series virtual impedance Zsj are
The Bode diagrams ofofparallel virtual impedance Z and
Zpj series virtualvirtual
impedance Z are shown
shown inThe Bode8.
Figure diagrams parallel
At the fundamental virtual impedance
frequency, andvirtual
thepjparallel series impedance impedance Zsj are
Zsjpj shows a high
in Figure 8. At the fundamental frequency, the parallel virtual impedance Z shows
shown in Figure 8. At the fundamental frequency, the parallel virtual impedance Zpj shows a high
pj a high impedance,
impedance, and the series virtual impedance Zsj displays a lowso impedance, so that the fundamental
and the series
impedance, virtual
and the series virtualZimpedance
impedance sj displays aZlow impedance, that the fundamental current
sj displays a low impedance, so that the fundamental
flows
current
intoflows
the into
grid. At the
the grid.frequency,
high At the highthe frequency,
parallel virtual the parallelZvirtual
impedance shows impedance
a low Zpj shows
impedance, and a low
the
current flows into the grid. At the high frequency, the parallel virtual pj impedance Zpj shows a low
impedance,
series and
virtual and
impedance,
the
impedance series
the series
virtual
Zsj displays impedance
virtuala high impedance,
impedance
Z sj displays a high impedance,
so that thea high-frequency
Zsj displays high impedance, harmonic so
so that
that
current
the
the
high-frequency
flows harmonic
into the
high-frequency parallel
harmonic current
virtual flows
flowsinto
impedance
current Zpjthe
into parallel
branch.
the virtualimpedance
parallel virtual impedance ZpjZbranch.
pj branch.
Magnitude (dB)
Magnitude (dB)
Phase (deg)
Phase (deg)
Figure 8. The Bode diagrams of parallel virtual impedance Zpj and series virtual impedance Zsj.
Figure 8. The
Figure 8. The Bode
Bode diagramsofofparallel
diagrams parallel virtual impedanceZZ
virtual impedance pj pjand
andseries virtual
series impedance
virtual Zsj . Zsj.
impedance
The Bode diagrams of the inverter self-impedance Zoj, parallel virtual impedance Zpj, grid
The Bode
equivalent diagramssgjof the grid
inverter self-impedance Zoj , parallel
Figurevirtual impedancewith Z pj , grid
The Bode impedance
diagrams Zof , the
and inverter impedance Zg are shown
self-impedance Zoj,inparallel 9.virtual
Combined impedance Figure Zpj5,
, grid
equivalent impedance
the grid equivalent Zsgj , andZgrid
impedance impedance Zg arethe
sgj is much lower than
shown in Figure 9.ZCombined
self-impedance with Figure 5,
oj and the parallel virtual
equivalent
the gridimpedance
equivalent Zsgj, and grid Zsgj impedance Zg are shown in Figure 9. Combined Zojcurrent with Figure 5,
impedance Zpj at theimpedance
fundamental is much thus
frequency, lower than
most the self-impedance
fundamental frequency and the parallel
flows into
the grid equivalent impedance Z sgj is much lower than the self-impedance Zoj and the parallel virtual
virtual branch withZrelatively
the gridimpedance pj at the fundamental
low impedance, frequency, thus most
which improves the fundamental
power qualityfrequency current
of grid-connected
impedance
flows Zpj the
into at the
grid fundamental
branch frequency,
with relatively lowthus most fundamental
impedance, which improves frequencythe powercurrent flows of into
current. At the high frequency, the parallel virtual impedance Zpj is much lower quality
than the
the grid branch with
grid-connected
self-impedance relatively
current.
Z lowhigh
At the impedance,
frequency, which improves
the parallel the impedance
virtual
oj and grid equivalent impedance Zsgj, thus most high-frequency harmonic
power quality Zpj is ofmuch
grid-connected
lower
current
current.
than At
flowsthe
into the high frequency,
self-impedance
the parallel Zoj and
virtual gridtheequivalent
impedance parallel virtual
impedance
Zpj branch impedance
Zsgj , thus
with relatively low Zpj high-frequency
most is much
impedance, andlower than the
harmonic
it effectively
current
self-impedance flows Z into
and the parallel
grid virtual
impedance
suppresses the oscillation of the paralleled multi-inverter system.
oj Z Z
pj branch
sgj , thus with
most relatively low
high-frequency impedance,
harmonic and it
current
effectively suppresses the oscillation of the paralleled multi-inverter system.
flows into the parallel virtual impedance Zpj branch with relatively low impedance, and it effectively
suppresses the oscillation of the paralleled multi-inverter system.
Energies 2018, 11, 3441 10 of 20
Figure 9.
Figure The Bode diagrams of the inverter self-impedance ZZojoj,, parallel virtual impedanceZZ
impedance pj,, grid
grid
Figure9.9.The
TheBode
Bodediagrams
diagrams ofof the
the inverter
inverter self-impedance parallel virtual
self-impedance Zoj, parallel virtual impedance Zpjpj
, grid
impedance Zsgj,,and
equivalent impedance Zgg. .
and gridimpedance
sgj, and grid
equivalent impedanceZZsgj grid impedance Z Zg.
4. Contrast Analysis of System Stability under Weak Grid Condition
4.4.Contrast
ContrastAnalysis
Analysisof ofSystem
System Stability
Stability under
under Weak Grid Condition
Weak Grid Condition
In order to compare and analyze the system stability, different conditions of adding parallel-series
In order
order toto compare and analyze the system stability,
stability, different
different conditions
conditions of of adding
adding
virtualInimpedances compare
are shown and
in analyze
Table 1. Parallel virtual impedance Zpj = r1 and series virtual
parallel-series
parallel-seriesvirtual impedances are shown in Table 1. Parallel
Parallel virtual
virtual impedanceZZpjpj==rr1 1and andseries
series
impedance Zsj virtual impedances
= r2 , abbreviated asare shown
case I. Parallel virtual impedance Zimpedance
pj = r 1 /G N and series virtual
virtual
virtual impedance
impedance ZZsjsj == rr22,, abbreviated
abbreviated as
as case I. Parallel virtual
virtual impedance
impedance ZZpjpj == rr11/G
/GNN and
andseries
series
impedance Zsj = r2 GN , abbreviated as case II (proposed control method).
virtual
virtualimpedance
impedanceZZsjsj==rr22GGNN,, abbreviated
abbreviated as case II (proposed control
control method).
method).
Table 1. Different conditions of adding parallel-series virtual impedances.
Table
Table1.1.Different
Different conditions
conditions of adding parallel-series
parallel-series virtual
virtual impedances.
impedances.
Number Parallel Virtual Impedance Zpj Series Virtual Impedance Zsj Control Method
Number
Number Parallel
ParallelVirtual
Virtual Impedance
Impedance Zpj pj Series Virtual Impedance
Impedance ZZsjsj Control
ControlMethod
Method
III rrr111 rr222 case
caseIII
case
II r1 /GN r2 GN case II
IIII rr11/G
/GNN r22GGNN case
caseIIII
To
ToToverify
verifythat
verify thatthe
that thesystem
the systemsatisfies
system satisfies the
satisfies the precondition
the precondition of usingusing the
using the Nyquist
the Nyquistcriterion,
Nyquist criterion,the
criterion, theopen-loop
the open-loop
open-loop
Bode
Bode diagrams
diagrams and
and closed-loop
closed-loop
Bode diagrams and closed-loop pole-zero pole-zero
pole-zero diagrams
diagrams of the
of equivalent
the equivalent coefficient of current source
equivalent coefficient of currentsource
coefficient of current Gj eq
source
for single
GGjeqjeqfor inverter
forsingle in
single inverter different
inverter in cases
in different
different casesare shown in Figure 10.
cases are shown in Figure 10. From Figure
10. From
From Figure 10a,b,
Figure 10a,b,it can be
10a,b,ititcan seen
canbe that
beseen
seen
the
thatgain
that the margin
thegain (GM)(GM)
gainmargin
margin and the
(GM) and
andphase margin
the phase
the phase (PM) (PM)
margin are greater than 0,
are greater
greater no right
than
than 0, nohalf-plane
0, no right pole exists,
right half-plane
half-plane pole
pole
and the single-inverter system is in a stable state. Therefore, the system
exists, and the single-inverter system is in a stable state. Therefore, the system satisfies the
exists, and the single-inverter system is in a stable state. Therefore, satisfies
the the
system precondition
satisfies of
the
using
preconditionthe Nyquist
precondition of criterion.
ofusing
using the The stability
the Nyquist
Nyquist condition
criterion.
criterion. of the paralleled
The stability
The stability condition multi-inverter
condition of
of the
the paralleled
paralleled system is that the
multi-inverter
multi-inverter
system
Nyquist is that
curve the
of Nyquist
the impedancecurve of
ratiothe impedance
does not ratio
surround ( −
does
system is that the Nyquist curve of the impedance ratio does not surround (−1, j0).1,not
j0).surround (−1, j0).
Figure
Figure10.10. The
The open-loop
open-loop Bode diagrams and closed-loop
closed-loop pole-zero
pole-zero diagrams
diagrams of of the
the equivalent
equivalent
Figure 10. The open-loop Bode diagrams and closed-loop
differentpole-zero diagrams of theII.equivalent
coefficient
coefficientofofcurrent sourceGGj eq
currentsource single inverter
jeq for single inverter in different
in cases.(a)
cases. (a)Case
CaseI;I;(b)
(b)Case
Case II.
coefficient of current source Gjeq for single inverter in different cases. (a) Case I; (b) Case II.
Energies 2018, 11, 3441 11 of 20
Figure 11 shows the Nyquist Nyquist diagrams of impedance impedance ratio K for for the
the paralleled
paralleled multi-inverter
multi-inverter
system. From Figure 11a, when the number of inverters inverters is 2, 3, 4, 5, and 6, respectively, the Nyquist
surround ((−1,
curve does not surround −1, j0)
j0) in
in case
case I,I, and the system is in a stablestable state.
state. However, when the
number of
number of inverters
invertersisis7,7,8,8,9,9,10,
10,and
and11,11, respectively,
respectively, thethe Nyquist
Nyquist curves
curves all surround
all surround (−1,(−1,
j0), j0),
andand
the
the system
system is inisan
in an unstable
unstable state.
state. However,from
However, fromFigure
Figure11b,
11b,regardless
regardlessof ofthe
the number
number of inverters,
the Nyquist curves would never wrap around ((−1, −1,j0)
j0)in
in case
case II,
II, and
and the
the system
system is is in
in aa stable
stable state.
state.
Therefore, compared with case I, the the proposed
proposed oscillation
oscillation suppression method by two notch filters
can effectively restrain parallel inverters’ harmonic current from from flowing
flowing into
into the grid, and avoid the
oscillation phenomenon.
Figure
Figure 11. Nyquist diagrams
11. Nyquist diagrams of
of impedance
impedance ratio K for
ratio K for the
the paralleled
paralleled multi-inverter
multi-inverter system.
system. (a)
(a) Case
Case I;
I;
(b) Case II.
(b) Case II.
5. Simulation Verification
To verify the correctness of theoretical analysis, the simulation model of a paralleled
multi-inverter system was built based on Figure 1. Several cases including two and seven inverters
Energies 2018, 11, 3441 12 of 20
5. Simulation Verification
To verify the correctness of theoretical analysis, the simulation model of a paralleled multi-inverter
system was built based on Figure 1. Several cases including two and seven inverters were simulated by
PSIM (Powersim Inc., Rockville, MD, USA) simulation. The system parameters are shown in Table 2.
When two inverters were connected to a weak grid in case I and case II, the root mean square
(RMS) value of the reference grid-side inductor current irefm for each grid-connected inverter increased
from 18.75 A to 37.5 A at 0.405 s. Therefore, the RMS of the reference grid-connected current igref
increased from 37.5 A to 75 A for the two parallel inverters. Simulation results of the grid-connected
current with two parallel inverters are shown in Table 3.
Table 2. System parameters.
Parameter/Unit Value Parameter/Unit Value

DC voltage Udc /V 720 Switching frequency f s /kHz 10
Grid phase voltage Ug /V 220 Proportional coefficient kp 2.1
Amplitude of triangular carrier Utri /V 1 Resonance gain ki1 175
Inverter-side inductor L1 /mH 2.2 Cut-off angular frequency ω c /rad/s 6.28
Parasitic resistance of L1 RL1 /Ω 0.002 Fundamental angular frequency ω o /rad/s 314
Filter capacitor C1 /µF 10 Fundamental frequency f o /Hz 50
Grid-side inductor L2 /mH 0.8 Quality factor Q 0.5
Parasitic resistance of L2 RL2 /Ω 0.001 Proportional coefficient r1 5
Grid inductor Lg /mH 2 Proportional coefficient r2 15
Table 3. Simulation results of grid-connected current with two parallel inverters. THD: total
harmonic distortion.
Before Reference Current Transient After Reference Current Transient

Number Case Resonance Point and Resonance Point and
THD THD
Resonance Peak Resonance Peak
I 6.07% 25th harmonic 1.77 A 3.16% 25th harmonic 1.81 A
2
II 3.27% 39th harmonic 0.88 A 1.51% 39th harmonic 0.79 A
In case I, the transient simulation waveforms of PCC voltage uPCC and grid-connected current ig
are depicted in Figure 12a,b. It can be seen that the distortion rate of PCC voltage uPCC is 1.57%.
The grid-connected current ig and corresponding spectrum from 0.30 s to 0.36 s are shown in
Figure 12c,d, which describes the situation before the reference current surges. The distortion rate of
grid-connected current ig is 6.07%, the resonance point is near the 25th harmonic (1250 Hz), and the
resonance peak is 1.77 A. The grid-connected current ig and corresponding spectrum after the reference
current suddenly increases are shown in Figure 12e,f. The distortion rate of grid-connected current ig
decreases to 3.16%, the resonance point is still near the 25th harmonic (1250 Hz), and the resonance
peak value is 1.81 A. At this time, the major high-frequency harmonics of the grid-connected current
are 25th harmonics.
In case II, the transient simulation waveforms of PCC voltage uPCC and the grid-connected current
ig are depicted in Figure 13a,b. From Figure 13a, the distortion rate of PCC voltage uPCC is 0.82%.
The grid-connected current ig and corresponding spectrum before the reference current surges are
shown in Figure 13c,d. The distortion rate of the grid-connected current ig is 3.27%, the resonance point
is near the 39th harmonic (1950 Hz), and the resonance peak is 0.88 A. The grid-connected current ig
and corresponding spectrum after the reference current suddenly increases are shown in Figure 13e,f.
The distortion rate of the grid-connected current ig decreases to 1.51%, the resonance point is still
near the 39th harmonic (1950 Hz), and the resonance peak value is 0.79 A. Meanwhile, the power
quality of the grid-connected current can be significantly improved, and the resonance phenomenon
has obviously decreased.
Energies 2018, 11, 3441 13 of 20
When seven inverters are connected to a weak grid in cases I and II, the RMS of reference
grid-side inductor current irefm for each grid-connected inverter is 10.71 A. Therefore, the RMS of the
reference grid-connected current igref is 75 A for seven parallel inverters. The simulation waveforms
and spectrograms of PCC voltage uPCC and the grid-connected current ig are shown in Figures 14
and 15. The simulation results of the grid-connected current with seven parallel inverters are shown in
15. The simulation results of the grid-connected current with seven parallel inverters are shown in
Table 4. In case I, the system is in an unstable state. The resonance phenomenon is obvious. The reason
Table 4. In case I, the system is in an unstable state. The resonance phenomenon is obvious. The
is that the high-frequency harmonic current frequency is equal to or close to the parallel resonance
reason is that the high-frequency harmonic current frequency is equal to or close to the parallel
frequency
resonance of self-impedance,
frequency resulting inresulting
of self-impedance, a parallelinresonance
a parallelor quasi-resonance
resonance of the impedance
or quasi-resonance of the
network. The grid-connected current is still severely polluted by the impedance
impedance network. The grid-connected current is still severely polluted by the impedance coupling interactions
coupling
between
interactionsthe between
invertersthe
andinverters
the grid.andIn case II, theIndistortion
the grid. rate
case II, the of the grid-connected
distortion current ig is
rate of the grid-connected
3.38%,
currenttheig isresonance
3.38%, thepoint is near
resonance the 39th
point is nearharmonic
the 39th (1950 Hz), (1950
harmonic and the resonance
Hz), peak is 1.51
and the resonance peak A.
Due to the sufficient resistive damping introduced to the impedance network,
is 1.51 A. Due to the sufficient resistive damping introduced to the impedance network, the systemthe system can operate
stably. Therefore,
can operate stably.case II can effectively
Therefore, case II canimprove the improve
effectively power quality of thequality
the power grid-connected current and
of the grid-connected
suppress
current and thesuppress
oscillation
theofoscillation
the paralleled
of the multi-inverter system.
paralleled multi-inverter system.
Figure 12. 12.Simulation

Simulationwaveforms
waveforms of point
of ofpoint
commonof coupling
common (PCC) voltage
coupling uPCC and
(PCC) grid-connected
voltage uPCC and
current ig in case
grid-connected I during
current ig inreference current
case I during transient
reference (two transient
current parallel inverters).
(two parallel(a) PCC (a)uPCC
voltage
inverters). PCC ;
(b) grid-connected
voltage current ig ; (c) grid-connected
uPCC; (b) grid-connected current ig (before);
current ig; (c) grid-connected (d) spectrogram
current of spectrogram
ig (before); (d) grid-connectedof
current ig (before);
grid-connected (e) grid-connected
current ig (before); (e) current ig (after); (f)current
grid-connected spectrogram of grid-connected
ig (after); current
(f) spectrogram of
igrid-connected
g (after). current ig (after).
Energies 2018, 11, 3441 14 of 20
100 two parallel inverters reference current transient
50 iga
ga/A igc
gc/A igb
gb/A
-50
-100 THD=3.27% THD=1.51%
0.36 0.38 0.4 0.42 0.44 0.46

Time (s)
(b)
Figure 13. Simulation

Figure 13. Simulation waveforms
waveforms of of PCC voltage uuPCC
PCC voltage PCC and the grid-connected
PCC current iigg in case
grid-connected current case II
II
during reference current transient (two parallel inverters). (a) PCC
during reference current transient (two parallel inverters). (a) PCC voltage uPCC voltage u ; (b) grid-connected
PCC; (b) grid-connected
PCC
current igigg;;(c)
current (c)grid-connected
grid-connectedcurrent
current ig (before); (d) spectrogram
igg (before); of grid-connected
(d) spectrogram current icurrent
of grid-connected g (before);
igg
(e) grid-connected
(before); current igcurrent
(e) grid-connected (after); i(f) spectrogram
gg (after); of grid-connected
(f) spectrogram current igcurrent
of grid-connected (after). igg (after).
Table 4. Simulation results of grid-connected current with seven parallel inverters.

Table 4. Simulation results of grid-connected current with seven parallel inverters.
Number Case THD Resonance Point and Resonance Peak
Number Case THD Resonance Point and Resonance Peak
I I (unstable)
(unstable) (unstable)
(unstable)
77
II II
3.38% 3.38% 39th harmonic
39th harmonic 1.51 A 1.51 A
200
100 iga
ga/A igc
gc/A igb
gb/A
-100
-200 unstable seven parallel inverters

0 0.05 0.1 0.15 0.2 0.25 0.3
Time (s)
(b)
Figure 14. Simulation

Figure 14. Simulation waveforms
waveforms of of PCC voltage uuPCC
PCC voltage grid-connected current iigg in case I
and the grid-connected
PCC and
PCC
(seven
(seven parallel
parallel inverters).
inverters). (a)
(a) PCC voltage uuPCC
PCC voltage PCC
PCC ; ; (b)
(b) grid-connected
grid-connected current iiggg. .
current
Energies 2018, 11, 3441 15 of 20
Figure Simulation
15.15.
Figure Simulationwaveforms
waveformsof ofPCC voltage uuPCC
PCCvoltage andthe
PCC and thegrid-connected
grid-connectedcurrentcurrent ig iin
g in case
case II II
Figure
(seven
(seven 15. Simulation
parallel
parallel waveforms
inverters).
inverters). (a) PCCof
(a)PCC PCC voltage
voltage
voltage uuPCC uPCC
PCC;; (b)
(b) and the grid-connected
grid-connected
grid-connected currentig;igcurrent
current ig in case of
; (c)spectrogram
(c) spectrogram II of
(seven parallel current
inverters).
grid-connected
grid-connected current igi.g. (a) PCC voltage uPCC; (b) grid-connected current ig; (c) spectrogram of
grid-connected current ig.
6. Experimental
6. Experimental Verification
Verification
6. Experimental Verification
ToTo
verify
verify thethevalidity
validityofofsimulation
simulation analysis,
analysis, a hardware-in-the-loop experimentalplatform
hardware-in-the-loop experimental platform
based
based To
on onverify
a real
a real the validity
controller
controller of simulation
TMS320F2812
TMS320F2812 analysis,
digital signal
digital a hardware-in-the-loop
processor
signal (DSP)
processor (Texas
(DSP) experimental
Instruments,
(Texas Instruments,platform
Inc, Dallas,
Inc,
TX,based
USA)on
Dallas, a real
and
TX, aUSA) controller
real-time TMS320F2812
and alaboratory
real-time (RT-LAB)digital
laboratory signal processor
(Opal-RT
(RT-LAB) (Opal-RT(DSP)
Technologies, (Texas Instruments,
Montreal,
Technologies, QC, Canada)
Montreal, Inc,
QC,was
Dallas,
Canada) TX,
was USA)
built and as
[26], a real-time
shown in laboratory
Figure 16a. (RT-LAB)
The system (Opal-RT
parameters Technologies,
are shown Montreal,
in Table 2. QC,
The
built [26], as shown in Figure 16a. The system parameters are shown in Table 2. The hardware-in-
Canada) was built [26],experimental
hardware-in-the-loop as shown in Figure 16a.mainly
The system parameters are shown in TableOP5700
2. The
the-loop experimental platform mainly platform
includes an RT-LAB includes
simulatoran OP5700
RT-LAB simulator
(Opal-RT Technologies,
hardware-in-the-loop
(Opal-RTQC, Technologies, experimental
Montreal, platform
QC, Canada), mainly includes
real controllers an RT-LAB
of grid-connected simulator
inverters, OP5700
a host
Montreal, Canada), real controllers of grid-connected inverters, a host computer as a real-time
(Opal-RT
computer Technologies,
as a real-time Montreal,
control QC, Canada),
interface, and anreal controllers as
oscilloscope, of shown
grid-connected
in Figure inverters,
16b. The amain
host
control interface, and an oscilloscope, as shown in Figure 16b. The main circuit model of the
computer
circuitwas as
model a real-time
of the system control
was interface,
established and an
in the oscilloscope,
hostwas
computer, as shown
which in
was Figure
loadedWhen16b. The
to the the main
OP5700.
system
circuit
established
model of the
in the
system
host
wasthe
computer,
established
which
in the host
loadedwhich
computer,
to thewas OP5700.
loaded
model
to thetoOP5700.
wasWhen the
running, model was
the OP5700 running,
sent thethe OP5700
analog sent
signals the analog
(voltages signals
and (voltages
currents) and
to the currents)
real controllers the real
through
When the model was running, OP5700 sent the analog signals (voltages
controllers through the input/output (I/O) ports in real time. After data processing, the real and currents) to the real
thecontrollers
input/output through(I/O)the
ports in real time.
input/output After
(I/O) datainprocessing,
ports realtotime. the realdata
After controllers transmitted the
controllers transmitted the digital control signals (pulses) OP5700 through processing,
the I/O portsthe in real
real
digital control
controllers signals
transmitted (pulses) to OP5700
theindigital controlthrough
signals the I/O ports in
(pulses) toprocessreal time.
OP5700tothrough The data interacted
thenormal
I/O ports in
in realreal
time. The data interacted real time during the simulation ensure the operation
time during
time. The the simulation
data interacted process
in real to ensure
time during the
the normal
simulation operation
process of
to the system.
ensure the The measurement
normal operation
of the system. The measurement information during the simulation were converted into analog
information
of during the
the system. simulation were converted into the
analog signals were
through the I/Ointo ports, which
signals throughThe measurement
the I/O ports, which information during
could be observed simulation
on the oscilloscope. converted analog
could be observed on the oscilloscope.
signals through the I/O ports, which could be observed on the oscilloscope.
Figure 16. Experimental platform for paralleled multi-inverter system based on hardware-in-loop
Figure 16.16.Experimental
Figure
simulation. Experimental platform
(a) Whole; (b)platform for
for paralleled
Structure. multi-invertersystem
paralleled multi-inverter systembased
basedononhardware-in-loop
hardware-in-loop
simulation. (a) Whole; (b) Structure.
simulation. (a) Whole; (b) Structure.
Energies 2018, 11, 3441 16 of 20

When two inverters operated in parallel, as in case I and case II, the RMS values of reference for
the grid-side
When inductor current
two inverters irefm for
operated in each grid-connected
parallel, inverter
as in case I and case II,increased
the RMS from
values18.75 A to 37.5for
of reference A at
0.405
the s. Therefore,
grid-side the RMS
inductor of reference
current grid-connected
irefm for each grid-connected current igrefincreased
inverter increasedfromfrom18.75
37.5 A A to
to37.5
75 AAfor
two parallel inverters. The transient experimental waveforms and spectrograms
at 0.405 s. Therefore, the RMS of reference grid-connected current igref increased from 37.5 A to 75 in case I and case
A II
arefor
depicted in Figures 17 and 18. Experimental results of the grid-connected
two parallel inverters. The transient experimental waveforms and spectrograms in case I and current with two parallel
inverters
case II are shown
depictedin in
Table 5. Before
Figures 17 and the18.
reference currentresults
Experimental surged ofin the Figure 17b, the distortion
grid-connected current withrate of
thetwo
grid-connected current
parallel inverters are ishown
g was 7.01%,
in Tablethe5. resonance
Before the point was
reference near
current the 25th
surged harmonic
in Figure (1250
17b, Hz),
the
and the resonance
distortion rate of peak was 2.66 A. After
the grid-connected currenttheigreference
was 7.01%, current suddenlypoint
the resonance increased
was near in Figure
the 25th17c,
theharmonic
distortion (1250
rate Hz),
of theand the resonancecurrent
grid-connected peak was 2.66 A. After
ig decreased to 4.22%,the reference current
the resonance suddenly
point was still
increased
near the 25thinharmonic
Figure 17c, the Hz),
(1250 distortion
and the rate of the grid-connected
resonance peak value was current ig decreased
2.89 A. to 4.22%, the
Thus, the grid-connected
resonance point was still near the 25th harmonic (1250 Hz), and the resonance
current apparently contained high-frequency ripples, and it is obvious that the major harmonics peak value was 2.89are
A. Thus, the
25th harmonics. grid-connected current apparently contained high-frequency ripples, and it is obvious
that the major
Before harmonics current
the reference are 25th surged
harmonics.in Figure 18b, the distortion rate of the grid-connected
Before the reference current
current ig was 4.82%, the resonance point was surged in Figure
near the18b, theharmonic
39th distortion(1950 rate of theand
Hz), grid-connected
the resonance
current ig was 4.82%, the resonance point was near the 39th harmonic (1950 Hz), and the resonance
peak was 2.06 A. After the reference current suddenly increased in Figure 18c, the distortion rate of
peak was 2.06 A. After the reference current suddenly increased in Figure 18c, the distortion rate of
grid-connected current ig decreased to 2.79%, the resonance point was still near the 39th harmonic
grid-connected current ig decreased to 2.79%, the resonance point was still near the 39th harmonic
(1950 Hz), and the resonance peak value was 2.04 A. Meanwhile, the power quality of grid-connected
(1950 Hz), and the resonance peak value was 2.04 A. Meanwhile, the power quality of
current can be significantly improved, and the resonance phenomenon has obviously decreased.
grid-connected current can be significantly improved, and the resonance phenomenon has
Therefore, the distortion rate was smaller after the reference current surge in the same case. Moreover,
obviously decreased. Therefore, the distortion rate was smaller after the reference current surge in
before and after
the same case. the reference
Moreover, current
before andsurged,
after thethe distortion
reference rate surged,
current in case II thewas less than
distortion ratecase I. II
in case
was less than case I.
Table 5. Experimental results of the grid-connected current with two parallel inverters.
Table 5. Experimental results of the grid-connected current with two parallel inverters.
Number Case
Resonance Point and Resonance Point
Number Case Resonance
THDPoint and Resonance Resonance Point and Resonance
THD
THD Resonance PeakTHD and Resonance Peak
Peak Peak
I I
7.01% 7.01%
25th harmonic 25th harmonic
2.66 A 2.664.22%
A 25th harmonic
4.22% 25th harmonic 2.89 A2.89 A
2 2 II 4.82% 39th harmonic 2.062.79%
A 2.79% 39th harmonic
39th harmonic
II 4.82% 39th harmonic 2.06 A 2.04 A2.04 A
Figure
Figure Experimental
17.17. waveforms
Experimental waveformsof PCC voltage
of PCC uPCCuPCC
voltage andandgrid-connected current
grid-connected ig inigcase
current I during
in case I
reference
during current
reference transient
current (two parallel
transient (twoinverters). (a) PCCuvoltage
(a) PCC voltage
parallel inverters). PCC andugrid-connected current ig ;
PCC and grid-connected
(b)current
grid-connected current ig and
ig; (b) grid-connected spectrogram
current (before); (c) (before);
ig and spectrogram grid-connected current ig and
(c) grid-connected spectrogram
current ig and
(after).
spectrogram (after).
Energies 2018, 11, 3441 17 of 20
18. Experimental waveforms of PCC

Figure 18. voltage uuPCC
PCC voltage and grid-connected
PCC and grid-connected current
current igg in case II
during reference current transient
transient (two
(two parallel
parallel inverters).
inverters). (a)
(a) PCC voltageuuPCC
PCC voltage PCC and grid-connected
current ig ; (b) grid-connected current ig and spectrogram (before); (c) grid-connected

grid-connected current iigg and
spectrogram (after).
When
When sevenseveninverters
invertersoperate
operate in parallel
in parallel in cases I andIcase
in cases andII,case
the RMS
II, theof RMS
the reference grid-side
of the reference
inductor
grid-sidecurrent
inductor irefm for each
current irefmgrid-connected inverter was
for each grid-connected 10.71 A.was
inverter Therefore,
10.71 A.the RMS of the
Therefore, thereference
RMS of
grid-connected current
the reference grid-connected i was 75 A for the seven parallel inverters. The experimental
gref current igref was 75 A for the seven parallel inverters. The experimental waveforms
and spectrograms
waveforms in cases I and
and spectrograms case III and
in cases are shown in Figures
case II are shown 19 and 20. 19
in Figures The experimental
and results of
20. The experimental
the
results of the grid-connected current with the seven parallel inverters are shown in Table 6. seen
grid-connected current with the seven parallel inverters are shown in Table 6. As can be in
As can
Figure 19, the system is in an unstable state. The resonance phenomenon
be seen in Figure 19, the system is in an unstable state. The resonance phenomenon is obvious. Theis obvious. The reason is that
the high-frequency
reason harmonic currentharmonic
is that the high-frequency frequencycurrent
is equalfrequency
to or closeistoequal
the parallel resonance
to or close to the frequency
parallel
of self-impedance, resulting in a parallel resonance or quasi-resonance
resonance frequency of self-impedance, resulting in a parallel resonance or quasi-resonance of the of the impedance network.
The grid-connected
impedance network.current is still severelycurrent
The grid-connected polluted by the
is still impedance
severely pollutedcoupling
by the interactions between
impedance coupling
the invertersbetween
interactions and the grid. From Figure
the inverters and the20, grid.
it canFrom
be seen that 20,
Figure the itdistortion
can be seenratethat
of the
thegrid-connected
distortion rate
current i is 4.36%,
g the resonance point is near the 39th harmonic (1950
current ig is 4.36%, the resonance point is near the 39th harmonic Hz), and the resonance
(1950 peak
Hz),
is
and2.99 A. resonance
the Due to the peaksufficient
is 2.99resistive
A. Due damping
to theintroduced
sufficient to the impedance
resistive dampingnetwork,
introducedthe system
to the
can operate network,
impedance stably. Therefore,
the system casecan
II can effectively
operate stably.improve the power
Therefore, case IIquality of the grid-connected
can effectively improve the
current and suppress the oscillation of the paralleled multi-inverter
power quality of the grid-connected current and suppress the oscillation of the paralleled system.
multi-inverter system.
Table 6. Experimental results of the grid-connected current with seven parallel inverters.
Table 6. Experimental
Number resultsCase
THD current with
Resonance seven
Point and parallel
Resonanceinverters.
Peak
Number Case THD
I Resonance Point and Resonance
(unstable) (unstable) Peak
7
I II
(unstable) 4.36% 39th harmonic 2.99 A
(unstable)
7
II 4.36% 39th harmonic 2.99 A
Energies 2018,
Energies 11,11,
2018, 3441
x FOR PEER REVIEW 18 18
of of
20 20
Figure 19. Experimental waveforms of PCC voltage uPCC and the grid-connected current ig in case I
Figure 19. Experimental
Experimental waveforms
waveforms of
of PCC voltage uuPCC
PCC voltage grid-connected current igg in case I
and the grid-connected
PCC and
(seven parallel inverters).
(seven parallel inverters).
Figure Experimental
20.20.
Figure Experimentalwaveforms
waveformsofofPCC voltageuPCC
PCCvoltage uPCCand
andthe
thegrid-connected current iigg in
grid-connected current in case
case II
Figure 20.parallel
II (seven Experimental waveforms
inverters). (a) PCC of PCC voltage
voltage u uPCC
and and the grid-connected
grid-connected current i ; current
(b) ig in case II
grid-connected
(seven parallel inverters). (a) PCC voltage PCCuPCC and grid-connected current ig; (b) grid-connected
g
(seven
current parallel
ig and inverters). (a) PCC voltage uPCC and grid-connected current ig; (b) grid-connected
spectrogram.
current ig and spectrogram.
current ig and spectrogram.
7. Conclusions
7. Conclusions
7. Conclusions
In this paper, the oscillation suppression method by two notch filters is proposed to realize the
In this paper, the oscillation suppression method by two notch filters is proposed to realize the
In impedances
virtual this paper, the andoscillation
increase the suppression
whole system method by twoThe
damping. notch filters is proposed
implementation form to realize
of the the
virtual
virtual impedances and increase the whole system damping. The implementation form of the virtual
virtual impedances
impedances and increase
is presented the wholePCC
by the proposed system damping.
voltage The implementation
feedforward and grid-sideform of thecurrent
inductor virtual
impedances is presented by the proposed PCC voltage feedforward and grid-side inductor current
impedances
feedback with is presented
two notch by the proposed
filters. PCC voltage
The feedforward path feedforward
of PCC voltage andwithgrid-side inductor
the notch filtercurrent
equals
feedback with two notch filters. The feedforward path of PCC voltage with the notch filter equals to
feedback
to a virtualwith two notch
impedance infilters.
parallel The feedforward
with path ofimpedance,
inverter output PCC voltage with is
which the notch filter
designed equals
to show to
low
a virtual impedance in parallel with inverter output impedance, which is designed to show low
aimpedance
virtual impedance in parallel
at the harmonic with frequency.
oscillation inverter output impedance,
By doing which is designed
so, most high-frequency to show
harmonic low
current
impedance at the harmonic oscillation frequency. By doing so, most high-frequency harmonic
impedance
will flow into at the
the parallel
harmonic oscillation
virtual impedancefrequency. Byand
branch, doing so, most suppresses
it effectively high-frequency harmonic
the oscillation.
current will flow into the parallel virtual impedance branch, and it effectively suppresses the
current
Meanwhile, will the
flow into thepath
feedback parallel
of thevirtual
grid-sideimpedance
inductor branch, and ittheeffectively
current with notch filter suppresses
is equivalent the
oscillation. Meanwhile, the feedback path of the grid-side inductor current with the notch filter is
oscillation.
to a virtual Meanwhile,
impedance in theseries
feedback
with path of the
inverter grid-side
output inductor
impedance, current
which with thetonotch
is designed filter
display lowis
equivalent to a virtual impedance in series with inverter output impedance, which is designed to
equivalent
impedance to a virtual
at the impedance
fundamental in series
frequency. Thiswith
way,inverter
most ofoutput impedance,
the fundamental which iscurrent
frequency designed to
flows
display low impedance at the fundamental frequency. This way, most of the fundamental frequency
display
into the low
gridimpedance
branch withatrelatively
the fundamental frequency.
low impedance. In This way,itmost
addition, of the the
improves fundamental
power quality frequency
of the
current flows into the grid branch with relatively low impedance. In addition, it improves the power
current flows into
grid-connected the grid
current. branch
Finally, with relatively
simulation low impedance.
and experimental resultsInare
addition,
provided it improves
to verify the thevalidity
power
quality of the grid-connected current. Finally, simulation and experimental results are provided to
quality of thecontrol
of proposed grid-connected
method. current. Finally, simulation and experimental results are provided to
verify the validity of proposed control method.
verify the validity of proposed control method.
Author Contributions: L.Y. and Y.C. provided the original idea for this paper. L.Y., H.W., A.L. and K.H. organized
theAuthor Contributions:
manuscript and attended L.Y.the
and Y.C. provided
discussions when the original
analysis andidea for thiswere
verification paper. L.Y., out.
carried H.W.,
AllA.L. and K.H.
the authors
Author Contributions: L.Y. and Y.C. provided the original idea for this paper. L.Y., H.W., A.L. and K.H.
gave comments
organized the and suggestions
manuscript and on the writing
attended and descriptions
the discussions of the manuscript.
when analysis and verification were carried out. All the
organized the manuscript and attended the discussions when analysis and verification were carried out. All the
authors gave
This comments
researchandand suggestions
was on the
supportedonby writing andKey
descriptions of the manuscript.
Funding:
authors gave comments suggestions thethe National
writing R&D Program
and descriptions of China under Grant No.
of the manuscript.
2017YFB0902000, and the Science and Technology Project of State Grid under Grant No. SGXJ0000KXJS1700841.
Funding: This research was supported by the National Key R&D Program of China under Grant No.
Funding: This research was supported by the National Key R&D Program of China under Grant No.
Conflicts Thethe
of Interest:and
2017YFB0902000, authors
Sciencedeclare no conflict Project
and Technology of interest.
of State Grid under Grant No. SGXJ0000KXJS1700841.
Acknowledgments: This work was supported by the National Key R&D Program of China under Grant No.
Acknowledgments: This work was supported by the National Key R&D Program of China under Grant No.
Energies 2018, 11, 3441 19 of 20
References
1. Yu, Y.; Konstantinou, G.; Hredzak, B.; Agelidis, V.G. Power balance optimization of cascaded H-bridge
multilevel converters for large-scale photovoltaic integration. IEEE Trans. Power Electron. 2016, 31, 1108–1120.
[CrossRef]
2. Guo, X.; Yang, Y.; Zhu, T. ESI: A novel three-phase inverter with leakage current attenuation for
transformerless PV systems. IEEE Trans. Ind. Electron. 2018, 65, 2967–2974. [CrossRef]
3. Bouloumpasis, I.; Vovos, P.; Georgakas, K.; Vovos, N.A. Current harmonics compensation in microgrids
exploiting the power electronics interfaces of renewable energy sources. Energies 2015, 8, 2295–2311.
[CrossRef]
4. Hany, M.H. Whale optimisation algorithm for automatic generation control of interconnected modern power
systems including renewable energy sources. IET Gener. Transm. Distrib. 2018, 12, 607–614.
5. Li, X.; Fang, J.; Tang, Y.; Wu, X. Robust design of LCL filters for single-current-loop-controlled grid-connected
power converters with unit PCC voltage feedforward. IEEE J. Emerg. Sel. Top. Power Electron. 2018, 6, 54–72.
[CrossRef]
6. Guo, X.; Yang, Y.; Wang, X. Advanced control of grid-connected current source converter under unbalanced
grid voltage conditions. IEEE Trans. Ind. Electron. 2018, 65, 9225–9233. [CrossRef]
7. Yang, Y.; Ye, Q.; Tung, L.J.; Greenleaf, M.; Li, H. Integrated size and energy management design of battery
storage to enhance grid integration of large-scale PV power plants. IEEE Trans. Ind. Electron. 2018, 65,
394–402. [CrossRef]
8. Rohner, S.; Bernet, S.; Hiller, M.; Sommer, R. Modulation, losses, and semiconductor requirements of modular
multilevel converters. IEEE Trans. Ind. Electron. 2010, 57, 2633–2642. [CrossRef]
9. Remon, D.; Cantarellas, A.M.; Mauricio, J.M.; Rodriguez, P. Power system stability analysis under increasing
penetration of photovoltaic power plants with synchronous power controllers. IET Renew. Power Gener. 2017,
11, 733–741. [CrossRef]
10. Malinowski, M.; Gopakumar, K.; Rodriguez, J.; Perez, M.A. A survey on cascaded multilevel inverters.
IEEE Trans. Ind. Electron. 2010, 57, 2197–2206. [CrossRef]
11. Amin, M.; Molinas, M. Small-signal stability assessment of power electronics based power systems:
A discussion of impedance- and eigenvalue-based methods. IEEE Trans. Ind. Appl. 2017, 53, 5014–5030.
[CrossRef]
12. Bakhshizadeh, M.K.; Yoon, C.; Hjerrild, J.; Bak, C.L.; Kocewiak, Ł.H.; Blaabjerg, F.; Hesselbæk, B.
The application of vector fitting to eigenvalue-based harmonic stability analysis. IEEE J. Emerg. Sel. Top.
Power Electron. 2017, 5, 1487–1498. [CrossRef]
13. Sun, J. Impedance-based stability criterion for grid-connected inverters. IEEE Trans. Power Electron. 2011, 26,
3075–3078. [CrossRef]
14. Yang, L.; Chen, Y.; Luo, A.; Chen, Z.; Zhou, L.; Zhou, X.; Wu, W.; Tan, W.; Guerrero, J.M. Effect of
phase-locked loop on small-signal perturbation modelling and stability analysis for three-phase LCL-type
inverter connected to weak grid. IET Renew. Power Gener. 2008. [CrossRef]
15. Xu, J.; Xie, S.; Tang, T. Improved control strategy with grid-voltage feedforward for LCL-filter-based inverter
connected to weak grid. IET Power Electron. 2014, 7, 2660–2671. [CrossRef]
16. Harnefors, L. Modeling of three-phase dynamic systems using complex transfer functions and transfer
matrices. IEEE Trans. Ind. Electron. 2007, 54, 2239–2248. [CrossRef]
17. Wang, X.; Blaabjerg, F.; Liserre, M.; Chen, Z.; He, J.; Li, Y. An Active Damper for Stabilizing
Power-Electronics-Based AC Systems. IEEE Trans. Power Electron. 2014, 29, 3318–3329. [CrossRef]
18. Yang, D.; Ruan, X.; Wu, H. A real-time computation method with dual sampling modes to improve the
current control performances of the LCL-type grid-connected inverter. IEEE Trans. Ind. Electron. 2015, 62,
4563–4572. [CrossRef]
19. Li, X.; Wu, X.; Geng, Y.; Yuan, X.; Xia, C.; Zhang, X. Wide damping region for LCL-type grid-connected
inverter with an improved capacitor-current-feedback method. IEEE Trans. Power Electron. 2015, 30,
5247–5259. [CrossRef]
20. Dannehl, J.; Fuchs, F.W.; Hansen, S.; Thogersen, P.B. Investigation of active damping approaches for PI-based
current control of grid-connected pulse width modulation inverters with LCL filters. IEEE Trans. Ind. Appl.
2010, 46, 1509–1517. [CrossRef]
Energies 2018, 11, 3441 20 of 20
21. Komurcugil, H.; Altin, N.; Ozdemir, S.; Sefa, I. Lyapunov-function and proportional-resonant-based control
strategy for single-phase grid-connected VSI with LCL filter. IEEE Trans. Ind. Electron. 2016, 63, 2838–2849.
[CrossRef]
22. Chen, Z.; Chen, Y.; Guerrero, J.M.; Kuang, H.; Huang, Y.; Zhou, L.; Luo, A. Generalized coupling resonance
modeling, analysis, and active damping of multi-parallel inverters in microgrid operating in grid-connected
mode. J. Mod. Power Syst. Clean Energy 2016, 4, 63–75. [CrossRef]
23. He, J.; Li, Y.; Bosnjak, D.; Harris, B. Investigation and active damping of multiple resonances in a
parallel-inverter-based microgrid. IEEE Trans. Power Electron. 2013, 28, 234–246. [CrossRef]
24. Zheng, C.; Zhou, L.; Xie, B.; Zhang, Q.; Li, H. A stabilizer for suppressing harmonic resonance in
multi-parallel inverter system. In Proceedings of the IEEE Transportation Electrification Conference and
Expo, (ITEC Asia-Pacific), Harbin, China, 1–6 August 2017.
25. Jeong, S.-G.; Park, M.-H. The analysis and compensation of dead-time effects in PWM inverters. IEEE Trans.
Ind. Electron. 1991, 38, 108–114. [CrossRef]
26. Shuai, Z.; Huang, W.; Shen, C.; Ge, J.; Shen, Z.J. Characteristics and restraining method of fast transient
inrush fault currents in synchronverters. IEEE Trans. Ind. Electron. 2017, 64, 7487–7497. [CrossRef]
© 2018 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).

Energies: Oscillation Suppression Method by Two Notch Filters For Parallel Inverters Under Weak Grid Conditions

Uploaded by

Copyright:

Available Formats

You might also like

Energies: Oscillation Suppression Method by Two Notch Filters For Parallel Inverters Under Weak Grid Conditions

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Energies: Oscillation Suppression Method by Two Notch Filters For Parallel Inverters Under Weak Grid Conditions

Uploaded by

Copyright:

Available Formats

energies

Energies 2018, 11, 3441; doi:10.3390/en11123441 www.mdpi.com/journal/energies

2. Oscillation Mechanism of Paralleled Multi-Inverter System

2.1. System Description

Figure 1. Structure of paralleled multi-inverter system.

3. Oscillation Suppression Method by Two Notch Filters for Parallel Inverters

3.1. Demand for the Inverter Output Impedance

Figure 3. Equivalent schematic diagram.

3.2. Oscillation Suppression Method

where r11 and r22 are

Energies 2018, 11, x FOR PEER REVIEW 6 of 20

idji = UUejej ·signi(iL1j ) (7)

Figure 5. Refined equivalent output impedance model.

Figure 6.6. Control block diagram

where f o is the fundamental frequency and Q is the s 2

Table 2. System parameters.

Parameter/Unit Value Parameter/Unit Value

Before Reference Current Transient After Reference Current Transient

Figure 12. 12.Simulation

100 two parallel inverters reference current transient

-100 THD=3.27% THD=1.51%

0.36 0.38 0.4 0.42 0.44 0.46

Figure 13. Simulation

Table 4. Simulation results of grid-connected current with seven parallel inverters.

-200 unstable seven parallel inverters

Figure 14. Simulation

Energies 2018, 11, x FOR PEER REVIEW 16 of 20

18. Experimental waveforms of PCC

current ig ; (b) grid-connected current ig and spectrogram (before); (c) grid-connected

You might also like